I thought about the last time I was flying. How was that experience? Provavelly, most people can think of some aspects that affected their previous lives, both positively and negatively, and will end up influencing their satisfaction.

When it comes to air travel, passenger satisfaction is a crucial metric for airlines. Knowing that your passages are satisfied is great, but knowing by reason of them we will be satisfied is even better. These insights can help an airline company understand where its strong points are and where we can improve, from the point of view of its customers. Let's see in more detail.

A study on customer satisfaction requested by passengers that evaluates their overall satisfaction, together with other aspects of their travel (for example, seat comfort, online reservation facility, leg room, departure and check-in delays, measured in minutes). This research confirmed that 54% two passengers were satisfied with their experience, which shows us that, in general, the majority of customers are satisfied with their experience.

Understand the reasons, two clients will be satisfied

It is very nice to know that most customers are satisfied with their travel experience. The next logical questions are: why are customers satisfied and what separates a satisfied customer from a neutral/dissatisfied customer? The search provides us with many predictors (>20) to investigate the given numbers (more than 100,000 lines). Starting from the Predictive Analysis menu, in the most recent version of Minitab Statistical Software, we can use CART® to quickly identify the main motivators of customer satisfaction.

CART®, or Classification and Regress Trees, is a decision tree algorithm used to help find important patterns and relationships in a variety of data. If you are wondering what problem you are facing in a binomial or multinomial categorical response, use Classification CART, and anything that has a continuous response with many categorical or continuous predictors should use Regress CART.

In this research, we are categorizing customers into two groups (satisfied or neutral/dissatisfied), so we will use CART Classification. The main idea behind CART is that we partition the predicted variables into different regions so that the dependent variable (also known as the alvo variable, which in this case is satisfiable) can be predicted more accurately. Minitab Statistical Software will automatically find the best decision tree for you and provide model statistics, so you can understand whether the model is useful.

To analyze these data, the parent model is quite large – or it is perfectly normal. Suppose you really want to concentrate and understand the main motivators of satisfaction. In this case, the graph of relative importance of variables can inform which predictors are the most important variables for the tree.

Flexibility to adjust an alternative model

Generally, CART trees can be very large. Viewing a smaller tree with similar information may be useful, especially if we plan to communicate our discoveries to other people. Fortunately, the most recent version of Minitab Statistical Software contains an interactive model visualization that allows us to explore, visualize and examine alternative models in a convenient way.

Tree diagrams to help you understand the details

Are CART files very useful when you want to understand how they vary? important, but also makes it easier for anyone to explore the variations by visualizing the values that generate the divisions in the model. Using the above, airlines will probably not be very surprised to know that customers want good in-flight entertainment and a comfortable seat – but knowing that even the in-flight entertainment is not optimal, we can keep passengers satisfied with a comfortable seat – é some important information.

The CART is a useful tool for your analysis toolbox, because it does not require many assumptions and can be completed quickly. If you have given that you do not analyze because predictive analysis and machine learning seem very intimidating for you – experimente or CART – which is even more easy in the most recent version of Minitab Statistical Software.

SOURCE: Minitab Blog

We are Software.com.br, Official Representative of Minitab Engageno Brazil and also a reference in technology solutions for the corporate world in Latin America. Count on our consultants specialized in Software Licensing, Cybersecurity, DevOps, Infrastructure and Data Analytics.

See more about Minitab on our site: Software.com.br